birds
Kurk, C.D.
Citation
Kurk, C. D. (2008, May 27). The bill of evolution : trophic adaptations in anseriform birds. Retrieved from https://hdl.handle.net/1887/12867
Version: Corrected Publisher’s Version
License: Licence agreement concerning inclusion of doctoral thesis in the Institutional Repository of the University of Leiden
Downloaded from: https://hdl.handle.net/1887/12867
Note: To cite this publication please use the final published version (if applicable).
Crani wildfow
C
ial geom wl (Aves: A
a
Chapte
etric mo Anatidae adaptatio
er 2
orphome e) indicat ons
trics of
te trophi ic
In cra of wh spe in Cra trie ge lan co rel len len nu Wh an the na fee A c ad bil as
a number of vert anial shape. The d
resource use, an hich natural selec ecialised feeding anatid birds is re anial shape of se ed to identify the ometric morpho ndmarks. Principa
mponents that m lative height of th ngths of pterygoi ngth and width o mber of allomet hile the cranial ch
d the reaction fo e bite and pulling rrow bill for an e eding, which requ comparison of gr
aptation to grazi l dimensions only
well.
tebrate taxa trop design of the tro nd in turn resourc ction can act. Wil g niches and in th eflected in cranial veral grazing and e characters that
metric approach al component an may be related to he neurocranium ds and palatines f the bill and wid ric shape change haracters indicat orces in the joints g forces required efficient transfer uires a large pum razing species wit ng reflects evolu y, while the more
Summary
phic specialization phic system dete ce use directly or ldfowl (Anatidae) is study we inves l design.
d filter-feeding an are related to di
was used to ana alysis of these da o grazing. The firs m, the position of . The second com dth of the cranium es.
ted by the PC ana s with the upper
during grazing. E of forces and is in mp capacity and t
thin the Anatidae utionary history. R e basal clades sh
ns have led to dis ermines the effici r indirectly influe ) have diverged i stigated whether
natid species was ifferences in feed alyse 33 three-dim
ata describes two st component des the craniofacial mponent describe
m. A third compo
alysis seem to be bill, the bill chara Efficient grazing r ncompatible with herefore a long a e suggests that th Recent clades po
ow large differen
stinct differences iency and bound nces fitness, upo nto a number of r trophic specializ
s examined and w ding habits. A mensional crania o independent scribes co-varian hinge, and the re es co-variation of onent indicates a
related to bite fo acters are related requires a short h efficient filter- and broad bill.
he degree of ossess modificatio
nces in cranial sh s in
aries on
zation
we l ce of elative f
orce d to
ons in ape
Fo ve e.g Go (W 20 an op pa up Th mo spe filt an is c ind Eff oro thr to be cu suc the ge tra Du up be cap Th on dim Ca al., sys fai ad the lev
raging performan rtebrate taxa, tro g., in bats (for rev ordon, 1999), rod Wainwrightet al., 05a). The design d the limits of pe perate. For foragi
tterns of resourc pon which natura e feeding mecha orphological mod
ecialized feeding ter-feeding ances cestor exploiting considered a seco dependently with ficiency of grazing opharyngeal epid rough repeated r 20 Hz (Kooloose ak. This foraging shion-like thicken cks water into th e tongue (so-calle
nerated. In many ansported over th uring these move pper bill (Van der comes very ineff pacity during filte e mechanisms of the shape of the mensions are rela kenberghe et al., , 2002; Wainwrig stem of most ver rly simple lever s vantage of a mus e bite force mom ver lengths (e.g. S
nce and cranial m ophic specializati view see Van Cak dents (Courantet 2004) and seed- of systems dete erformance set th
ng this means th ce use. In turn, re l selection can ac anisms in wildfow dification and per g niches ranging f
stor (Olson and F g submerged vege
ondarily derived hin the anatid cla g and filter-feedi dermal morpholo rapid opening an et al., 1989) to ge
mechanism requ nings of the tong
e oral cavity. Foo ed ‘under-tongue y grazing species, he tongue during ments food is ret Leeuwet al., 200 ficient (e.g., malla er-feeding (see c f grazing and filte e skull. In several ated to the amou , 2002; Pérez-Bar ght et al., 2004; v rtebrates produce system. The theo
scle as the ratio o ment arm (out-lev
Schenk and Wain
Introduction
morphology are f ons have led to d kenberghe et al., t al., 1997), turtle cracking birds (va rmines how effic he ultimate boun at design and pe esource use direc ct (Arnold, 1983) wl (Anatidae) prov
rformance. Anati from piscivory to Feduccia, 1980; Zw
etation (Van der feeding mechan ade.
ng are believed t ogy (see also chap
d closing movem enerate a waterfl
uires a bald palat gue, which acts as
od that is filtered e’ transport) so t , however, grass g a series of for/b tained by small s 03). Without the ard), while grazin hapter 4 and 5).
er-feeding in wild vertebrate taxa unt of bite force g
rbería and Gordo van der Meij, 200
es bite force by t ory of force transm
of the muscle for ver). Bite force ab nwright, 2001; W
n
functionally linke distinct differenc 2002), ungulates es (Herrel et al., 2 an der Meij, 2004 ciently behaviors ndaries within wh rformance deter ctly or indirectly i
.
vide an example id species have d herbivory, presu weers and Vande Leeuwet al., 200 ism and has evol
to impose conflic pter 3). Filter-fee ments of the jaws ow with food pa tal surface, and w
s a piston within d out of the wate
hat a continuous or seeds filtered backward movem spines on the inne
se spines the tra ng species have a
dfowl may also po it has been show generated during on, 1999; Couran 04; Herrel et al., 2 the jaw adductor mission defines t rce moment arm bility will thus im
estneat, 1990). I
d. In a number o es in cranial shap s (Pérez-Barbería 2002), fishes
4; Herrel et al., can be performe hich an animal m rmine individual nfluences fitness
of the link betwe diverged into umably from eith
en Berge, 1997) o 03). Terrestrial gr
ved several time
cting demands on eding is performe with a frequency rticles through th well developed
the opened bill a r is transported a s waterflow can b from the water ments of the tong er surface of the nsport of grass a very low pump
ose opposite dem wn that cranial
g feeding (Van t et al., 1997; He 2005a). The feedi muscles acting o the mechanical
(in-lever) relativ prove at short ou n grazing anserifo
f pe, a and
ed, ust s,
een er a or an razing es
n the ed
y up he and along be
are ue.
mands
rrel et ing on a
e to ut-
orms,
a f he the ele spe inc fre clo Th rel mu are she mo 19 allo an Pa We be ge Alt dif he Ph So 2.1 an spe (Ch du oc Nu in An
forceful closure o ad and neck are e bill. Furthermo evate the upper b
ecies may benefi crease their pum equencies up to 2 osing movements e differences in f lated to both sku usculature. To inv e matched by cra
eldgeese and duc orphological dive
97) of cranial fea ow size and shap d offer powerful rsons et al., 2003 e also investigate tween the two s ometry has evolv ternatively, differ fferent foraging h rbivory among g ylogenetic analy renson et al., 199 1). From a comm
d swans (Cygnus ecialized grazing hloephaga, Alopo cks (Anas), the w casionally is the o ummi, 1993; Com
our study as leas natinae and a num
of the bill is neces drawn backward re, the backward bill, which is mov it from relatively p volume capacit 20 Hz and the dra s may be conside force regimes act ull geometry (rela vestigate whethe anial morphology cks) are studied.
ersity between an atures using land pe to be consider techniques for s 3; Adams and Ro ed whether there
ubfamilies Anatin ved along similar rences in skull ge habits, difference roups, or the tim ses (Madsen et a 99; Livezey, 1997 on ancestor two s) (Anserinae) and species have div ochen, Neochen a wigeon (Anaspen
omnivorous mall mbs and Fredricks st specialized gra mber of Anas spe
ssary to hold gras ds, the grass will s d movement of th veable with respe
short bills, filter- ty. However, filte ag forces generat erable.
ting on the bill du ative positions of
er the functional y several groups o
To investigate th natid groups, we mark-based mor red independentl studying variation
hlf, 2000).
e is significant ov nae and Anserina r pathways to me eometry may refl es in constraints t me of independen al., 1988; Sraml e 7a) suggest the fo
groups originate d the duck-like bi verged within the and Cyanochen), nelope). Not spec lard (Anasplatyrh son, 1996; Drilling zer. Straining spe ecies (shovelers)
ss firmly in the bi snap off, rather t he head will resu ect to the neuroc -feeders may ben er-feeding is perf ted during high v
uring grazing and joints and muscl demands of filte of anatid birds (g he extent and nat
constructed a m rphometric meth ly, preserve geom n in form (Rohlf a
erlap in skull geo ae, which would eet the mechanic ect selection pre that influence th nt evolution.
t al., 1996; Donn ollowing successi ed: the true geese
irds (Anatinae). S e Anatinae clade.
and last, within t cialized, but know hynchos) (Arzel a g et al., 2002). Th ecialists are foun
are included in o
ill so that when t than be pulled ou
lt in forces that cranium. While gr
nefit from large b formed at high
elocity opening a
d filter-feeding m les) and the size er-feeding and gra geese, swans,
ture of cranial orphospace (Foo ods. These meth metric informatio and Marcus, 1993
ometry of herbivo suggest that skul al demands of gr essures related to
e evolution of ne-Goussé et al., 2
on of events (fig e (Anser and Bra Subsequently,
First the sheldge the genus of dab wn to graze
and Elmberg, 200 he mallard is inclu
d only within the our analyses.
the ut of
razing bills to and
ay be of jaw azing
ote, ods on,
3;
ores ll razing.
o
2002;
ure nta) eese bbling 04;
uded e
Fig Ma cha
Sp Th pri (N (W ge Ch rep rhy (An are
A
Anatidae
A
gure2.1. Phylogene adsen et al., 1988;
aracterization of th
ecimens
is study is based ivate collection o ational Museum Wageningen Unive
ese (Anser and B loephaga), and b presented by 4 sh ynchotis). The stu nasplatyrhyncho e listed in table 2
Anas ( Anatinae
sheldge
swans
nserinae true ge
etic relationships (c Donne-Goussé et a he anatid species e
Ma
on 150 skulls of of G. Niklaus (Pad of Natural Histor ersity). Grazing sp Branta), and all bu by the Eurasian w
hoveler species ( udy further comp os). The exact num 2.1.
dabbling ducks)
eese
eese
compiled from Live al., 2002; Johnson
xamined in this stu
aterialsandMe
adult anseriform dingbüttel, Germa
ry, Leiden, The N pecialists are rep ut one species of wigeon (Anaspen Anasclypeata, A prised all 7 specie
mber of specime
filter-feeders
omnivore herbivore
herbivores
herbivores herbivores herbivores
ezey, 1996a, 1997b and Sorenson, 199 udy.
ethods
m specimens. Spec any), the collecti Netherlands) and presented by all 1
f sheldgeese (Alo nelope). Filter-fee Anasplatalea, An es of swans (Cygn
ns per species (s
A.clypeata
A.smithii
A.rhynchotis
A.platalea A.platyrhyncho
A.penelope
Chloephaga
Alopochen
Neochen
Cyanochen
Cygnus
Anser
Branta
b; Sraml et al., 1996 99) and trophic
cimens belong to on of Naturalis W. van Gestel 15 species of true
pochen, Neochen eding specialists a assmithii, and A nus) and the mal ubspecies are po
os
6;
o the
e n and are
nas
lard
ooled)
Tab
A
B
C
A
ble2.1. Identificati Scientific Anser
A.cygnoid A.fabalis A.brachyr A.anser
A.albifron A.erythro A.indicus A.canagic A.caerule A.rossii
ranta
B.bernicla B.leucops B.canade B.sandvic B.ruficolli Cygnus
C.atratus
C.melanc C.olor
C.buccina C.columb C.columb C.cygnus
Cyanochen Alopochen Neochenj Chloephag Chloephag Chloephag Chloephag Anas
A.penelop A.platyrhy A.smithii
A.platalea
A.rhyncho A.clypeat
ion and number of name
des
rhynchus
ns
opus
cus
escens
a
sis
nsis
censis
is
oryphus
ator
ianus
ianusbewickii
ncyanoptera
naegyptiaca
jubata
gamelanoptera
gapicta
gapoliocephala
garubidiceps
pe
ynchos
a
otis
ta
f Anseriform specie Com grey geese
Swan g
Bean go
Pink-fo Greylag Greate
Lesser W
Bar-hea Empero
Snow g
Ross' go
black geese
Brent g
Barnac Canada Hawaiia Red-bre Swans
Black sw
Black-n
Mute s
Trumpe Whistli Bewick Whoop Sheldgeese Blue-w Egyptia Orinoco Andean Magella Ashy-he Ruddy- dabbling ducks Europe Mallard
Cape sh
Red sho
Austral norther
es studied.
mmon name
goose oose
oted goose g Goose
r White-fronted go White-fronted goo aded goose or goose goose
oose
goose le goose a goose
an goose (Nene) easted goose
wan necked swan
wan eter swan ng swan k's swan per swan
inged goose an goose
o goose n goose
an (Upland) goose eaded goose -headed goose
ean wigeon d
hoveler oveler
ian shoveler rn shoveler
# 43
oose ose
29
30
25
23 1 9 3 6 4 3 3 2 8 4
10 6 6 3 4
4 5 4 3 1 4 9
1 2 4 3 6 6 3
4 3 5 4 5 2
Lan Th ge for an sys the dig rec sho are sym mo Th of ste 50 Co fro
Fig rot B a lon par
ndmarks
e variation of an ometric morpho rm of the skull, to
d Fisher, 1962), a stem. Essential to e landmarks (Boo gitised at intersec corded in the for own in figure 2.2 e only unilateral mmetry, some co orphometric ana
ex, y, and zcoor each skull. A sku epwise around th cm and perpend oolpix 950) was m om a dorsal viewp
gure2.3. Experimen tation starting from and C, and part of t ngitudinal axis, B: X
rt.
seriform cranial metry. Landmark o encompass line and to include lan o the geometric m okstein, 1991) be
ctions of bony str rm of three-dime 2 and listed in tab
in order to reduc ontralateral landm
lyses (see below) rdinates of each l ll was clamped a he longitudinal ax dicular to the lon made. All photogr point, a total of 8
ntal set-up. Eight d m a dorsal view of t the graph paper. A XYZ frame attached
morphology is an ks for this analysi ear measurement ndmarks related morphometric te etween specimen ructures. For eac nsional (3D) coor ble 2.2 (see next p ce the number of
marks were inclu ).
landmark were c t the orbits in a r xis with intervals gitudinal axis of t raphs had a resol 8 pictures were ta
digital pictures wer
the anatid skull. Ea : knob of rotating d d to stationary part
nalysed using lan is were chosen to ts used in a previ
to lever lengths echnique is the bi ns. Most of the la ch specimen, 33 l rdinate data. The page). Although f variables by tak uded for orientat
ollected from a s rotating device (f of 30 degrees. A the skull, a digita ution of 1200 x 1 aken for each sku
re taken from a 30 ch picture included device to rotate an t, C: XYZ frame atta
dmark-based o cover the geom ious study (Good
of the jaw muscl iological homolog
ndmarks were andmarks were e landmarks used
most of the land ing advantage of ion of the skulls i
series of photogr figure 2.3) and ro At a fixed distance
al photograph (Ni 1600 pixels. Start
ull.
degree gradual d the total skull, pa natid skull around it ached to rotating
metric dman
e gy of
d are marks f
in the
aphs otated
e of ikon ting
art ts
Fig cor
Tab me
N 1 2 3 4 5 6 7 8 9 1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2
gure2.2. Position a rrespond to table 2
ble2.2. Number an easured contralate
Number of landmar
/2c /4c /6c /7c 0/10c 1 2/12c
3 4 5
6 7
8 9
0 1 2 3 4/24c 5 6
and number of land 2.2. A: lateral view,
nd description of la rally. Terminology
rk Descripti tip of the largest w connecti connecti connecti articulati most ros most cau articulati tip posto most ros most cau articulati tip of orb medial co lateral co condylus condylus articulati highest p caudalm lateralmo posterior ventralm base of p middle o
dmarks used to rep , B: ventral view (m
andmarks. See also according to (Baum
ion of landmark e maxilla
width maxilla just ca on jugale with max on palatinum with on vomer with ma ion frontal nasal hi stral point of lining udal point of lining ion palatinum with orbital process stral point of basipt udal point of basipt ion pterygoid with bital process of qua ondyle of processu ondyle of processu s medialis quadrati s lateralis quadratii ion jugale with qua point cranium
ost point on the cr ost point on crista rmost point on the most point of the pa postorbital process of occipital condyle
A
present anatid skul modified from Dela
o figure 2.2. Points melet al., 1993).
audal from maxilla xilla
maxilla xilla
inge with maxilla of the orbit of the orbit h pterygoid
terygoid surface on terygoid surface on
quadrate adrate
us oticus quadratii s oticus quadratii i
i adrate
rista nuchalis transv nuchalis transvers e prominentia cere aroccipital process s
l morphology. Num cour, 1964).
marked with c wer
ry nail
n skull n skull
versa a bellaris
B
mbers
re also
Cu dig un of an we sca Th firs ad ind rot co wa (N eff rot dif the to va of co No ha nu ph ph Th bo cle
Ge Th spe 19 an Pro va dif rem sca lan co
stom software (R gitise and subseq
intended transla a XYZ metal coor d the tips of a XY ere digitised in ea aling factor for th e 3D coordinates st estimate of its ding a random co dividual measure
tated to the sam mbined standard as used as a cost elder and Mead fectively gave the tation plane (x do fference with the
e plane in which converge to accu lue) was estimate the algorithm, th ordinates.
ot all landmarks w s to be digitised mber of times a otographs, occas otographs.
e overall standar oth x and y direct
early higher and s
eometricmorpho e three-dimensio ecimens were an 91; Rohlf and Ma d shape to be co ocrustes superim riables from the fferences in spec moved. Each spe aled to the avera ndmarks to the ce
mmon orientatio
R. G. Bout) writte quently compute ation of the skull rdinate frame fix YZ metal coordina ach of the 8 pictu he images.
s of each landma unknown third c omponent to a se ement. The series
e (lateral) orienta d deviation over a function that wa simplex method;
e same result as s oes not change u e measured x, y v they were measu urate values (95%
ed from a data se he final set of val were visible in all in at least two ph landmark is mea sionally coordina rd deviation after ion. However, fo slightly different
ometrics
onal set of 33 lan nalysed by geome arcus, 1993; Dryd nsidered as two mposition (GLS) (G
set of homologo imen position, or cimen is translat ge centroid size entroid of the co on that minimizes
en in Matlab (The the three coordi and rotation erro ed to the station ate frame fixed t ures. A piece of g
rk were reconstr coordinate was c
eries of 10 values s of pictures cont
ation after a corr all x, y and z mea as minimized with
; (Bunday, 1984)) starting with ran under the rotatio values after rotati
ured. The numbe
% of the coordina et with known va ues for each land photographs. To hotographs, altho sured. Most poin ates were estimat r convergence fo
r rotating points in x (0.24 mm) a
ndmarks (also cal etric morphomet den and Mardia, independent com Gower, 1975; Roh
us landmarks rec rientation, and si ted onto a comm (square root of t onfiguration), and
s the squared dif
e Mathworks Inc.
nates of each lan ors with respect t nary part of the d
o the rotating pa raph paper was u
ructed as follows hosen. A search s of the first estim taining the landm rection for the pr asurements in the h a steepest grad ) by adjusting the dom y and z-valu n scheme used) a ion of the initial c er of cycles requi ates less than 0.0 alues and varianc dmark was avera o estimate 3D-co ough the accurac nts of the skull w ted from 2 or mo or a stationary po the (pooled) sta nd y direction (0.
led configuration trics (Marcus et a 1998; Rohlf, 199 mponents. A gen hlf and Slice, 199 corded on each s ize have been ma
on centroid, then he sum of square d lastly each spec fferences betwee
., Natick) was use ndmark. To check to the camera, th evice (B in figure art (C in figure 2.3 used to calculate
. For each landm matrix was creat mate for each mark were then a
rojection angle. T e lateral rotation dient descent met
e z-value. This ues in the lateral
and minimizing t coordinate towa red for the algori 002 mm from the ce. After converg ged to estimate ordinates a landm cy increases with were visible in 3 o
ore than 4 oint was 0.1 mm i
ndard deviation w .17 mm).
ns) for all 150 al., 1996; Bookste
8) which allows s eralized least-sq 0) generates sha pecimen, after athematically
n all specimens a ed distances from cimen is rotated t en corresponding
ed to k for he tips e 2.3)
3) e the
mark a ted by
ll The
plane thod
he rds ithm
ir true ence its mark
the r 4
in was
ein, size
uares ape
are m all
to a g
lan Bo mo thr Th (Ha exc ne Mo ne sid pe exc va dis 12 po 25 wa
Sta Aft lan SP To (M of lan qu a s To co the sam tes an rec int nu for Wh va
ndmarks (Rohlf a ookstein (1991) a orphometric shap roughout the ana e Generalized Pr ammer et al., 200 clusively ipsilater gatively affect th oreover, the cons
urocranial eleme de either, which m
rformed on a sub cluded to avoid t riation between stribution of cran , 24) measured la oints (3, 9, 13, 15,
) were not used t ay as the landma
atisticalprocedu ter superimposit ndmark) and subj
SS10 (SPSS Inc. C test differences MANOVA). Differe univariate ANOV ndmarks). Althou estioned (Perneg study wide (type
reduce the dime mponent analysi e data were suita mpling adequacy sts (KMO = 0.80, alysis. The PCA w commended if va terpretation of th mber of variable r further analysis hile differences i riation correlated
nd Slice, 1990; A nd Rohlf (1990).
pe variables pres alysis.
rocrustes Superim 01) and custom w ral landmarks sho he fit of points on
struction of the a ents does not allo may also affect th bset of the landm the Pinocchio-eff
groups than neu nial landmarks a n
andmarks were i , 16, 17, 18, 19). T to determine the rks used for the G
ures
ion, transformed jected to statistic Chicago), unless s
between groups ences between co VA’s. This involve ugh the use of Bo ger, 1998), we als
I) error rate of p ensionality of the
s. Only ipsilatera able for factor an y tests and Bartle Bartlett: p = 0.00 was performed on ariances are very he factors, a varim es with high loadi s.
n geometric scale d with size (i.e. a
dams et al., 2004 Unlike linear dist serve the geomet
mposition was pe written software owed that the be n the less intensiv avian skull with it ow an even distri he fit of the landm marks. The three
fect (Walker, 200 rocranial landma number of media ncluded, as well The remaining la e best fit but tran
GPS.
d coordinates we cal analysis. Stati stated otherwise.
s a multivariate g oordinates of lan s a large number nferroni adjustm so calculated the
= 0.001
e data set of align al landmarks were nalyses, both the ett's test of spher 00) confirmed tha
n the correlation different (Quinn max rotation was ngs on a factor. T
e are removed d llometric compo
4). Details of this tance measurem try of the anatom
erformed using PA (R.G. Bout). GPS est fit for ipsilater
vely sampled con ts many fused an ibution of landma
marks. The GPS w landmarks (1, 2, 00), as bill length arks. To obtain th al (5, 26) and all b as half of the ips andmarks (8, 11, nslated, scaled an
re treated as var stical procedures .
eneral linear mo dmarks were ass r of univariate tes ment in our type o e Bonferroni-Holm
ned coordinates w e entered in the
Kaiser-Meyer-Ol ricity were perfor at the data were matrix. Standard n and Keough, 20
s performed. This The first three co
uring the GPS, as nents of shape) a
method are give ents, geometric mical structure
AST-software S with sets of
ral points may ntralateral side.
nd poorly delinea arks on the ipsila was therefore
2c) of the bill we showed much la he most even
bilaterally (4, 6, 7 ilateral measure 14, 20, 21, 22, 23 nd rotated in the
iables (3 for each s were performe
del was used sessed through a
sts (99 for 33 of analyses may b m test. This resul
we used principa PCA. To test whe kin measure of rmed. Both diagn
suitable for facto dization is
02). To simplify t s minimizes the omponents were
spects of shape are not. To check
en in
ted ateral
re rger 7, 10,
d 3, and
same
h d with
series be
ted in
al ether nostic
or the
used
k for
allo ce Th to pro dif the Fin dis co co 20 the
Re Aft wa AN ch Gr po co Bra va lar a r gro gro oth spe Dif spe dif the the oth Sh fol
ometric effects, t ntroid size were e scores of each determine signif oved to be signif ffered from one a
e shoveler specie nally, a morpholo stances between ordinates were a mputed on unwe 01). Both trees w ese landmarks w
elativelengthsof ter Procrustes fit as done by calcul NOVA's on groups
eck for differenc oups differ in sku ostorbital process
mpared to Anas anta species hav riation in skull wi rgest in Branta an relatively short sk oups. Both malla oup. In the wigeo her two skull wid ecialised filter-fe fferences in skull ecies, the palatin ffer among each
e other groups, b e two grazing An her Anas species eldgeese and Bra lowed by Anser,
the correlation b computed.
specimen on the ficant differences icant, post-hoc te another. For stat es to form a singl ogical distance tre
each pair of sku approximated by
eighted pair-grou were based on all as used in which
fcranialboneele tting and standar
ation the relative s with species me es between grou ull length and in s ses. Anser, Branta species, and a br e significantly wi idth at the occipi nd smallest in An kull, although not rd and wigeon ha on the width of th dth measures, bo
eding ducks, but length are not re ne is shorter than
other. Similarly, t but only differs si
as species, malla , but only the ma anta species poss Cygnus and Anas
between PC score e three principal c s among the anse ests on group me istical reasons, b e Anas group.
ee was estimated ll configurations
Procrustes resid up averages (UPG
l ipsilateral landm the landmarks o
Results
ementsinwildfo dizing skull size, e lengths of a num eans, followed by ups. The results a
skull width at the a, Cygnus and sh roader skull at th der skulls than A ital processes is l as and Cygnus. W t as short as the g ave a relatively w he hinge is simila th mallard and w t the differences
eflected in palati n in the other gro the pterygoid bo gnificantly from ard and wigeon, h
allard has a relati ses the relatively s species. Sheldg
es of sample mea components wer eriform groups. W eans were used t
oth wigeon and m
d using the matri of landmarks. Ta uals and two phe GMA) using PAST marks, for the sec of the bill were ex
wl
a first exploratio mber of bone ele y post-hoc tests, re listed in table e craniofacial join eldgeese species e craniofacial joi Anas at the posto ess than at more Within the Anas g grazing species in wide craniofacial
ar to the width in wigeon show larg
are small.
ne and pterygoid oups (except Bran one is longer in An
Cygnus and Anas have clearly longe ively short palatin y shortest bills am geese, Branta and
ans per species an re used in an ANO When the ANOVA
o assess which g mallard were add
x of Procrustes angent space
enetic trees were (Hammeret al., cond tree a subse xcluded.
n of shape differ ements.
were performed 2.3.
nt and at the s all have shorter
nt. Only Anser an rbital level. The e rostral levels an
group the wigeon n the other anati
hinge within the sheldgeese. For er values than th
d lengths. In Anse nta), which do no nserspecies than s species. Note, t er pterygoids tha ne.
mong all anatid gr d Anser species a
nd OVA A
roups ded to
e et of
rences d to
skulls nd nd is
n has d
Anas the he 4
er ot n in that an the
roups lso
ha he the hig lar Th spe of lar Ho co wit Or see
ve narrower bills ight is significant e Anas group, the gher) bills than th rger than in the B e quadrate is lon ecies, with Cygnu
the quadrate wit rgest in Cygnus, A owever, the diffe
mparisons show th foraging mech rbit size in Cygnus ems to have the
s tips compared t tly larger in Anser e mallard and esp he filter-feeding d Branta species.
nger in both true us taking an inter th the skull and t Anser and Branta rences in quadra significant differ hanism.
sspecies is signif largest orbit size
to Cygnus and the r and Branta spe pecially the wige ducks. In the wig
geese genera co rmediate position the two condyles a species and sma
te width are very rences, and there
ficantly smaller th of all anatid spe
e long-billed Ana cies than in all ot on have shorter eon relative bill l
mpared to sheld n. The distance b of the quadrato allest in sheldgee y small at both ar e seems to be no
han in all other g cies measured.
as species. Maxill ther groups. With
and narrower (b length is only slig
geese and Anas between the two
mandibular joint ese and Anas.
rticulations, few obvious relation
roups and the w a hin ut not ghtly
joints t are
ship
igeon
Table2.3. Mean lengths with standard deviations of skull elements of anseriform groups at average centroid size. To compare filter-feeders and grazers within the Anas group data for the northern shoveler and wigeon are given separately. When not stated otherwise all three coordinates are used for calculat Superscript numbers indicate significant differences between groups. Tamhane (T) instead of Bonferroni tests are used when Levene's test indicates inhomogeneity of variance. 1. Anser (n = 10) 2. Branta (n = 5) 3. Cygnus (n = 7) 4. sheldgeese (n = 7) 5. Anas (n = 6) northern shoveler skull length mid 6 - 23 5 - 23 58.11 ± 2.434, 5 59.15 ±1.545 60.99 ± 1.205 61.00 ± 0.78360.17 ± 1.305 57.22 ± 1.272,561.82 ± 1.071, 5 58.72 ± 1.505 66.91 ± 1.731, 2, 3, 4 61.35 ± 1.471, 3, 467.34 59.30 skull width 6z - 6c z 10z - 10c z 24z - 24c z 17.95 ± 1.083, 4, 5 34.97 ± 2.343, 5 23.66 ± 0.99 16.36 ± 1.135 35.09 ± 0.633, 5 25.02 ± 0.823, 5
14.88 ±1.161, 5 31.14 ± 1.351, 2 22.54 ± 0.702,4
14.35 ± 1.051, 5 32.97 ± 1.37 24.36 ± 0.633, 5
12.18 ± 1.581, 2, 3, 4 30.43 ± 1.771, 2 22.48 ± 0.852, 4
11.08 28.47 21.82 Maxilla length 1 - mid 6 57.83 ± 5.432, 4 49.12 ± 2.911, 3, 5 57.21 ± 2.902, 4 45.25±3.051, 3, 5 68.55 ± 9.692, 4 79.05 Maxilla width 2z - 2c z 17.69 ± 1.963, (5) 17.41 ± 0.833, 5 21.54 ± 0.991, 2, 4 16.28 ± 2.283, 5 31.64 ± 7.40(1), 2, 4 40.81 Maxilla height mid 6y – 5y15.79 ± 1.633, 4, 5 14.43 ± 0.6512.21 ±0.561 13.47 ± 1.451 12.85 ± 1.221 13.37 Palatine length 4 - 9 29.38 ± 1.993, 4, 5 30.89 ± 1.9334.43 ± 1.851 33.67 ± 2.851 33.84 ± 2.311 34.43 pterygoid length 9 - 13 14.20 ± 0.913, 5 13.23 ± 0.64 12.11 ± 2.051 12.44 ± 1.1311.88 ± 1.591 11.07 Basipterygoid 11 - 12 6.36 ± 0.616.33 ± 0.47 6.12 ± 0.316.73 ± 0.785 5.65 ± 0.295.78 quadrate length 16-18 15-17 14.32 ± 0.674, 5 13.86 ± 0.904, 514.18 ± 0.464 14.00 ± 0.374, 513.87 ± 0.66 13.21 ± 0.6012.90 ± 0.701, 2 12.67 ± 0.721, 213.00 ± 0.781 12.68 ± 0.581, 212.62 12.25 quadrate width 15 - 16 17 - 18 4.11 ± 0.455 5.05 ± 0.423.97 ± 0.32 4.88 ± 0.254.30 ± 0.325 5.06 ± 0.20 3.60 ± 0.48 4.53 ± 0.303.24 ± 0.541, 3 4.67 ± 0.484.27 4.95 orbit size 7 – 8 23.16 ± 1.403 24.55 ± 0.443 17.71 ± 0.761, 2, 4, 5 22.81 ± 1.463 23.02 ± 1.303 22.58
Dif Aft the lan 3D No sho Th on tes cri (p All ips the
fferencesinanat ter scaling and su e position of the ndmarks at the ve D-coordinates per
ot surprisingly, m ows significant d e comparison of e coordinate and sts 81 show a sig
terion p < 0.05, a
< 0.001) (table 2 except one coor silateral counterp e skull (points1,
tidskullshapeba uperimposition (f
coordinates of th entral side of the r group after sup ultivariate analys ifferences in hea
individual landm d contribute to th nificant differenc and 61 are signifi
.4).
rdinate of the con part. As expected
5, 23, 26) do not
asedonlandmar figure 2.4), large he bill and of the e skull show smal erimposition are sis of variance of ad shape between marks shows that
he shape differen ce between the g cantly different a ntralaterally mea d, the z-coordinat t differ among gro
rkcoordinatecon differences betw e dorsal part of th ler differences. I e listed.
f the coordinates n the five groups almost all landm nce between the groups according
according to the asured landmarks
tes of landmarks oups.
nfigurations
ween groups exist he cranium, while n table 2.4, the m
of each landmar s (p = 0.000).
marks differ in at l groups. Of the 9 to the conventio Bonferroni-Holm s behave like the
in the medial pla t in e the mean
rk least
9 onal m test
eir ane of
Fig con sho sku pro
gure2.4. Positions nfiguration per gro own). A: Upper par ull configurations, B ojection. *: Anser, {
of the cranial land oup after superimp rt: lateral view of s B: Upper part: vent {: Branta, : Cygn
marks and represe osition procedures kull, lower part: co tral view of skull, lo nus,: Sheldgeese,
entation of the mea s, the latter depicte orresponding proje
ower part: corresp , and +: Anas.
an landmark ed in scaled mm (n ection of superimpo
onding dorso-vent not
osed tral
A
B
Tab P-v AN
1X 1Y 1Z 2X 2cX 2Y 2cY 2Z 2cZ 3X 3Y 3Z 4X 4cX 4Y 4cY 4Z 4cZ 5X 5Y 5Z 6X 6cX 6Y 6cY 6Z 6cZ 7X 7cX 7Y 7cY 7Z 7cZ 8X 8Y 8Z 9X 9Y 9Z 10X 10c 10Y 10c 10Z 10c
ble2.4.Means and values indicate the NOVA, * indicate sig
Anser (n = 10) mean ± s.e.
77.69 ±1.94 -0.056 ±0.46 0.26 ±0.27
65.25 ±1.33 X 65.28 ±1.37 0.48 ±0.37 Y 0.41 ±0.37
-8.98 ±0.48 Z 8.71 ±0.24 31.97 ±0.40 0.19 ±0.38 9.76 ±0.14 31.97 ±0.56 X 32.24 ±0.53 1.21 ±0.31 Y 0.62 ±0.24
7.33 ±0.14 Z -7.42 ±0.15
23.64 ±0.18 -0.03 ±0.26 -0.02 ±0.07 22.05 ±0.35 X 22.23 ±0.30 -15.80 ±0.30 Y -15.83 ±0.33 8.95 ±0.19 Z -8.99 ±0.21
9.66 ±0.31 X 10.12 ±0.32 -20.18 ±0.37 Y -20.07 ±0.36 9.53 ±0.16 Z -9.25 ±0.20
-12.86 ±0.23 -16.91 ±0.53 13.69 ±0.33 2.88 ±0.31 3.26 ±0.14 3.96 ±0.17 X -3.48 ±0.22
cX -2.54 ±0.23 Y -4.35 ±0.34
cY -4.86±0.26 Z 17.63 ±0.33
cZ -17.32 ±0.43
d standard errors o probability that th gnificance accordin
) .
Branta (n = 5) mean ± s.e.
4 69.79 ±1.59 6 0.23 ±0.92 7 0.35 ±0.22 3 58.89 ±1.61 7 58.78 ±1.62 7 0.50 ±0.65 7 0.46 ±0.64 8 -8.43 ±0.26 4 8.986 ±0.18 0 31.97 ±0.49 8 0.09 ±0.38 4 9.34 ±0.18 6 32.00 ±0.40 3 32.31 ±0.39 1 0.75 ±0.26 4 0.25 ±0.23 4 7.51 ±0.15 5 -7.02 ±0.16 8 23.88 ±0.05 6 -0.70 ±0.23 7 0.08 ±0.11 5 23.05 ±0.34 0 23.39 ±0.28 0 -14.94 ±0.19 3 -15.30 ±0.16 9 8.52 ±0.31 1 -7.82 ±0.27 1 9.61 ±0.37 2 10.04 ±0.35 7 -19.64 ±0.60 6 -20.17 ±0.72 6 9.60 ±0.15 0 -8.99 ±0.25 3 -14.09 ±0.31 3 -15.56 ±0.42 3 14.46 ±0.29 1 1.37 ±0.46 4 1.86 ±0.40 7 3.86 ±0.14 2 -3.58 ±0.63 3 -2.39 ±0.70 4 -2.67 ±0.24 6 -3.18 ±0.30 3 17.65 ±0.11 3 -17.42 ±0.20
of landmark coordi he samples are stat ng to the Bonferon
Cygnus (n = 7) mean ± s.e.
83.77 ±1.08 -8.67 ±1.43 -0.04 ±0.26 72.28 ±1.17 71.82 ±1.20
-6.57 ±1.14 -6.54 ±1.14 -10.71 ±0.31 10.81 ±0.17 33.97 ±0.15 -0.01 ±0.30 8.22 ±0.08 34.04 ±0.15 34.13 ±0.18 -0.47 ±0.16 -1.11 ±0.18 7.17 ±0.17 -7.23 ±0.16 24.66±0.33 -1.93 ±0.18 -0.09 ±0.13 26.86 ±0.42 27.06 ±0.33 -14.09 ±0.26 -14.17 ±0.23 7.98 ±0.36 -6.89 ±0.22 5.34 ±0.26 5.36 ±0.28 -17.67 ±0.24 -17.60 ±0.36 8.98 ±0.18 -8.13 ±0.21 -11.65 ±0.15 -13.51 ±0.38 11.57 ±0.18 -0.15 ±0.62 2.15 ±0.22 4.28 ±0.15 -3.90 ±0.30 -3.82 ±0.26 -2.84 ±0.31 -3.36 ±0.28 15.66 ±0.23 -15.47 ±0.29
nates after superim tistically identical a
i-Holm test (p < 0.0 Sheldgeese
(n = 7) mean ± s.e.
A m 67.91 ±1.51 9
1.96 ±0.89 -1 0.38 ±0.14 58.95 ±1.04 8 58.89 ±1.11 8 1.63 ±0.69 - 1.60 ±0.67 - -8.41 ±0.45 -1 7.87 ±0.46 1 33.63 ±0.63 3
0.86 ±0.49 8.25 ±0.25 33.62 ±0.65 3 33.64 ±0.66 3
0.57 ±0.25 -0.16 ±0.30 -
7.21 ±0.19 -6.68 ±0.30 - 23.88 ±0.30 2 -0.79 ±0.32 -
0.17 ±0.05 25.74 ±0.29 2 25.85 ±0.25 2 -14.10 ±0.40 -1 -14.42 ±0.36 1
7.29 ±0.23 -7.05 ±0.24 -
6.65 ±0.45 6.95 ±0.36 -20.19 ±0.69 -1 -20.10 ±0.63 -1
8.73 ±0.18 -8.08 ±0.18 - -14.89 ±0.28 -1 -14.91 ±0.59 -1 13.97 ±0.44 1 0.19 ±0.54 - 2.10 ±0.22 3.75 ±0.20 -5.61 ±0.22 - -4.41 ±0.30 - -2.92 ±0.34 - -3.47 ±0.36 - 16.56 ±0.21 1 -16.38 ±0.31 -1
mposition for each according to a univ
001).
Anas ( n= 6) mean ± s.e. p-val 98.11 ±4.59 0.000 10.51 ±0.88 0.000 0.35 ±0.27 0.705 82.93 ±4.06 0.000 83.04 ±4.25 0.000 -7.93 ±0.82 0.000 -7.94 ±0.85 0.000 15.96 ±1.47 0.000 15.68 ±1.56 0.000 33.09±0.52 0.011 0.24 ±0.25 0.609 7.77 ±0.27 0.000 32.87 ±0.47 0.022 33.00 ±0.48 0.053 0.17 ±0.30 0.002 -0.80 ±0.25 0.000 6.90 ±0.16 0.172 -6.51 ±0.21 0.027 25.23 ±0.45 0.002 -0.80 ±0.28 0.002 0.11 ±0.06 0.193 29.66 ±0.67 0.000 29.70 ±0.62 0.000 13.61 ±0.27 0.000 13.70 ±0.28 0.000 5.99 ±0.14 0.000 -6.19 ±0.57 0.000 4.41 ±0.64 0.000 4.30 ±0.60 0.000 18.60 ±0.58 0.013 18.83 ±0.37 0.005 7.41 ±0.16 0.000 -7.10 ±0.36 0.000 17.84 ±0.19 0.000 14.55 ±0.39 0.000 11.54 ±0.44 0.000 -0.67 ±0.52 0.000 2.15 ±0.50 0.006 3.00 ±0.24 0.001 -6.65 ±0.28 0.000 -6.22 ±0.48 0.000 -3.97 ±0.35 0.005 -4.48 ±0.24 0.002 15.35 ±0.31 0.000 15.06 ±0.41 0.000
taxon.
ariate
lue 0*
0*
5 0*
0*
0*
0*
0*
0*
1
9 0*
2
3 2
0*
2 7
2
2
3 0*
0*
0*
0*
0*
0*
0*
0*
3
5
0*
0*
0*
0*
0*
0*
6
1*
0*
0*
5
2
0*
0*